基于EM方法的隐Markov软件可靠性模型

Hidden Markov Software Reliability Model with EM Method

针对单一软件可靠性模型不能准确描述软件失效行为、无法合理准确地评估预测出软件可靠性的问题,将变点分析引入软件可靠性建模,提出了一种基于隐Markov过程的软件可靠性模型。该模型采用隐变量来描述影响软件可靠性的多种因素,通过隐变量的状态变化刻画出软件过程中各种因素的变化情况,构建出隐Markov链软件可靠性模型,并采用EM算法进行求解,通过实例分析来验证其有效性。实验结果表明,隐Markov链软件可靠性模型具有较强的变点检测能力,并能显著提高软件可靠性拟合精度。

In view of the problem that single software reliability model doesn＇t precisely describe the failure behavior of the software, and doesn＇t accurately predict the software reliability, this paper studied a hidden Markov chain software reliability model incorporating the change point analysis. The formulation of the hidden Markov chain software reliabili- ty prediction approach involves a hidden state variable that indicates the regime change. This variable is specified to be detected by software failure data in each regime. The model parameters are estimated using expectation/maximization （EM） algorithm. Some numerical examples were performed based on some real software failure data sets. Experimental results show that the proposed framework to incorporate multiple change points for software reliability model has fairly accurate and efficient change-point detection capability, and can significantly improve software reliability fitting accuracy.